[0/2] btrfs: small improvement to leaf dump

Message

Qu Wenruo April 27, 2023, 12:16 p.m. UTC

The first patch is just constify the tree dump infrastructure, exposed
during the development.

The second patch is the main dish, inspired by a report that turns out
to be bitflip in extent tree.
However the leaf dump in dmesg is pretty large (100+ slots), needs to
manually search the leaf using the bytenr from the error messages.

Thus if we can output the slot number, it would be much easier to locate
the problem, just like what we did in tree-checker.

Qu Wenruo (2):
  btrfs: print-tree: accept const extent buffer pointer
  btrfs: improve leaf dump and error handling

 fs/btrfs/ctree.c       |   8 +--
 fs/btrfs/ctree.h       |   2 +-
 fs/btrfs/extent-tree.c | 125 +++++++++++++++++++----------------------
 fs/btrfs/print-tree.c  |  16 +++---
 fs/btrfs/print-tree.h  |   4 +-
 5 files changed, 73 insertions(+), 82 deletions(-)

Comments

David Sterba May 2, 2023, 3:34 p.m. UTC | #1

On Thu, Apr 27, 2023 at 08:16:26PM +0800, Qu Wenruo wrote:
> The first patch is just constify the tree dump infrastructure, exposed
> during the development.
> 
> The second patch is the main dish, inspired by a report that turns out
> to be bitflip in extent tree.
> However the leaf dump in dmesg is pretty large (100+ slots), needs to
> manually search the leaf using the bytenr from the error messages.
> 
> Thus if we can output the slot number, it would be much easier to locate
> the problem, just like what we did in tree-checker.
> 
> Qu Wenruo (2):
>   btrfs: print-tree: accept const extent buffer pointer
>   btrfs: improve leaf dump and error handling

Added to misc-next, thanks.